[SWDEV-509876] Remove buffer requirement from device counting service (#132)

* [SWDEV-509876] Remove buffer requirement from device counting service

No longer require a buffer to be given when setting up device counting
service. This is to reduce performance overhead in cases where immediate
return of counting samples is being used (synchronous mode).

* Missed file

* Update source/include/rocprofiler-sdk/device_counting_service.h

Co-authored-by: Madsen, Jonathan <Jonathan.Madsen@amd.com>

* Update source/lib/rocprofiler-sdk/counters/controller.cpp

Co-authored-by: Madsen, Jonathan <Jonathan.Madsen@amd.com>

* Update source/lib/rocprofiler-sdk/counters/device_counting.cpp

Co-authored-by: Madsen, Jonathan <Jonathan.Madsen@amd.com>

* Fixes for build

---------

Co-authored-by: Benjamin Welton <ben@amd.com>
Co-authored-by: Madsen, Jonathan <Jonathan.Madsen@amd.com>
Co-authored-by: Benjamin Welton <bewelton@amd.com>

[ROCm/rocprofiler-sdk commit: 0c4a56c6bb]

projects/rocprofiler-sdk/source/include/rocprofiler-sdk/device_counting_service.h

@@ -76,7 +76,9 @@ typedef void (*rocprofiler_device_counting_service_callback_t)(
  * @param [in] context_id context id
  * @param [in] buffer_id id of the buffer to use for the counting service. When
  * rocprofiler_sample_device_counting_service is called, counter data will be written
- * to this buffer.
+ * to this buffer. If the input buffer id is null (i.e. `rocprofiler_buffer_id_t{.handle = 0}`), the
+ * counter data will not be written to a buffer and will only be returned in the output_records of
+ * rocprofiler_sample_device_counting_service
  * @param [in] agent_id agent to configure profiling on.
  * @param [in] cb Callback called when the context is started for the tool to specify what
  * counters to collect (rocprofiler_profile_config_id_t).

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/controller.cpp

@@ -82,7 +82,8 @@ CounterController::configure_agent_collection(rocprofiler_context_id_t context_i
     // cannot coexist in the same context for now.
     if(ctx.pc_sampler) return ROCPROFILER_STATUS_ERROR_CONTEXT_CONFLICT;
 
-    if(!rocprofiler::buffer::get_buffer(buffer_id.handle))
+    if(!rocprofiler::buffer::get_buffer(buffer_id) &&
+       buffer_id != rocprofiler_buffer_id_t{.handle = 0})
     {
         return ROCPROFILER_STATUS_ERROR_BUFFER_NOT_FOUND;
     }

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/counters/device_counting.cpp

@@ -132,7 +132,7 @@ agent_async_handler(hsa_signal_value_t /*signal_v*/, void* data)
         *prof_config->agent, prof_config->required_special_counters, decoded_pkt);
 
     auto* buf = buffer::get_buffer(callback_data.buffer.handle);
-    if(!buf)
+    if(!buf && callback_data.buffer != rocprofiler_buffer_id_t{.handle = 0})
     {
         ROCP_FATAL << fmt::format("Buffer {} destroyed before record was written",
                                   callback_data.buffer.handle);
@@ -160,8 +160,9 @@ agent_async_handler(hsa_signal_value_t /*signal_v*/, void* data)
             {
                 callback_data.cached_counters->push_back(val);
             }
-            buf->emplace(
-                ROCPROFILER_BUFFER_CATEGORY_COUNTERS, ROCPROFILER_COUNTER_RECORD_VALUE, val);
+            if(buf)
+                buf->emplace(
+                    ROCPROFILER_BUFFER_CATEGORY_COUNTERS, ROCPROFILER_COUNTER_RECORD_VALUE, val);
         }
     }